Cool. Found one.
Unfortunately schedule_timeout() doesn't actually call del_timer_sync.
Rather odd. Perhaps the scheduler has swizzled the stack pointer or
something...
Could you please add some extra diags and rerun?
if (count == 0) {
printk( "del_timer_sync(%p): deadlock! Called from %p\n",
timer, __builtin_return_address(0));
printk("See
http://www.uow.edu.au/~andrewm/linux/deadlock.html\n");
+ show_stack(0);
+ printk("handler=%p\n", timer->function);
}
Then feed the extra stack addresses into gdb. Also, feed the 'handler='
address into gdb. This will help to identify where the deadlock is
occurring.
Thanks.
Pete Toscano wrote:
>
> hello,
>
> i was (am) using 2.4.0test1-ac13 to try to put an end to incessant
> lockups that i've been seeing since 2.3.99-pre7 days (maybe earlier, i
> don't quite recall). the problem is, ac13 has a bug in buffer.c that
> "make -j bzlilo" tends to trigger, so i just upgraded to ac14.
>
> upon booting ac14, everything looks good for a couple seconds, then the
> following message appears on the screen, looping until i power-cycle:
>
> del_timer_sync (c195bf4c): deadlock! Called from c0117868
> See http://www.uow.edu.au/~andrewm/linux/deadlock.html
>
> so, i checked out the mentioned page and followed the mentioned
> procedure. since i only had one address listed, the following is kind
> of sparce, but here it is:
>
> (gdb) x/10i 0xc0117868
> 0xc0117868 <schedule_timeout+128>: mov 0xc028bd4c,%eax
> 0xc011786d <schedule_timeout+133>: mov %esi,%ebx
> 0xc011786f <schedule_timeout+135>: sub %eax,%ebx
> 0xc0117871 <schedule_timeout+137>: xor %eax,%eax
> 0xc0117873 <schedule_timeout+139>: cmp $0xffffffff,%ebx
> 0xc0117876 <schedule_timeout+142>: cmovg %ebx,%eax
> 0xc0117879 <schedule_timeout+145>: lea 0xffffffe0(%ebp),%esp
> 0xc011787c <schedule_timeout+148>: pop %ebx
> 0xc011787d <schedule_timeout+149>: pop %esi
> 0xc011787e <schedule_timeout+150>: mov %ebp,%esp
>
> my machine is a dual p3-600 with 512m ram. asus p2b-d mobo.
>
> if there's any futher info i cna provide, please let me know.
>
> btw, i'm not subscribed to the linux-smp list, so please write me
> directly.
>
> thanks,
> pete
>
> --
> Pete Toscano p:[EMAIL PROTECTED] w:[EMAIL PROTECTED]
> GPG fingerprint: D8F5 A087 9A4C 56BB 8F78 B29C 1FF0 1BA7 9008 2736
>
> ------------------------------------------------------------------------
> Part 1.2Type: application/pgp-signature
--
-akpm-
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
